UAV-DualCog - Self-Relative Position: leaderboard
UAV-DualCog (2026) evaluates multimodal LLMs on self-state and environment-state reasoning of a UAV in simulated aerial scenes, with structured answers in instant mode. This board asks where the UAV is relative to a landmark from two or five aerial images; answer accuracy over 1,024 items in percent, higher is better.
Metric: Answer accuracy (%; 1,024 questions asking where the UAV is relative to a landmark from 2 or 5 aerial images; structured JSON answers on renders of simulated AerialVLN scenes, the discrete answer scored whatever the predicted box or interval; instant mode with explicit thinking disabled where a model allows it; higher is better). Source: arxiv.org. Saturation forecast: Around 2030. 36 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (Non-reasoning) | 38 | #49 (GPT-5.5) |
| 2 | GPT-5.4 (Non-reasoning) | 37.9 | #123 (GPT-5.4) |
| 3 | Claude Sonnet 4.6 | 37.7 | #161 |
| 4 | Kimi K2.6 (Non-reasoning) | 36.1 | #178 (Kimi K2.6) |
| 5 | GPT-5.3 Instant | 35.2 | #130 |
| 6 | Kimi K2.5 (Non-reasoning) | 34.3 | #256 (Kimi K2.5) |
| 7 | Gemini 3.1 Flash Lite (Non-reasoning) | 33.7 | #364 (Gemini 3.1 Flash Lite) |
| 8 | Qwen 3.7 Plus (Non-reasoning) | 33.3 | #207 (Qwen 3.7 Plus) |
| 9 | MiMo-V2.5 (Non-reasoning) | 33.1 | #354 (MiMo-V2.5) |
| 10 | Qwen 3.6 Plus (Non-reasoning) | 32.8 | #222 (Qwen 3.6 Plus) |
| 11 | Qwen 3.5 122B A10B (Non-reasoning) | 32.6 | #331 (Qwen 3.5 122B A10B) |
| 12 | Qwen 3.5 27B (Non-reasoning) | 31.6 | #359 (Qwen 3.5 27B) |
| 13 | Qwen 3.5 4B (Non-reasoning) | 30.9 | #1175 (Qwen 3.5 4B) |
| 14 | GLM-4.6V (Non-reasoning) | 30.8 | #701 (GLM-4.6V) |
| 15 | Qwen 3.5 35B A3B (Non-reasoning) | 29.5 | #482 (Qwen 3.5 35B A3B) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=uav-dualcog-self-relative-position · How It Works · Data refreshed daily, snapshot 2026-11-13.